My model throws weird output in predict function:
from tensorflow import keras
from tensorflow.keras import layers
import pandas as pd
data = pd.read_csv("/kaggle/input/water-potability/water_potability.csv")
wat_train = data.sample(frac=0.7, random_state=0)
wat_valid = data.drop(wat_train.index)
max_ = wat_train.max(axis=0)
min_ = wat_train.min(axis=0)
wat_train = (wat_train - min_) / (max_ - min_)
wat_valid = (wat_valid - min_) / (max_ - min_)
# Split features and target
X_train = wat_train.drop('Potability', axis=1)
X_valid = wat_valid.drop('Potability', axis=1)
y_train = wat_train['Potability']
y_valid = wat_valid['Potability']
print(X_train.shape)
model = keras.Sequential([
layers.BatchNormalization(input_shape=[9]),
layers.Dense(4, activation="relu"),
layers.BatchNormalization(),
layers.Dropout(0.3),
layers.Dense(4, activation="relu"),
layers.BatchNormalization(),
layers.Dropout(0.3),
layers.Dense(1, activation="sigmoid")
])
model.compile(
optimizer='adam',
loss='binary_crossentropy',
metrics=['binary_accuracy'],
)
early_stopping = keras.callbacks.EarlyStopping(
patience=10,
min_delta=0.001,
restore_best_weights=True,
)
model.fit(X_train, y_train,
validation_data=(X_valid, y_valid),
batch_size=512,
epochs=1000,
callbacks=[early_stopping],
)
predictions = model.predict(X_valid)
print(predictions)
It prints:
[[nan]
[nan]
[nan]
[nan]
[nan]
[nan]
[nan]
[nan]]
I am using Water Quality Dataset on kaggle: https://www.kaggle.com/datasets/adityakadiwal/water-potability
The output layer is sigmoid function. I want to return between 0 -> 1 But this returns unlimited number of [nan]
